[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: How to replace missing values in panel data
also expenditure tends to be highly skewed, so you might be better off
imputing log expenditure, which is approximately normal excepty for
somewhat heavier tails. Keep in mind that you create a lot of problems
my imputation: your standard errors do not account for the fact that
you've estimated expenditure, rather than have hard data; and you have
an explicit measurement error in your imputed variable, that leads to
all sorts of biases and, again, smallish standard errors.
Stata journal had a special issue on measurement error models; I
believe this was SJ 3 (4).
You may also want to consider multiple imputation methods that correct
for some of the above problems.
* For searches and help try: